KR101443569B1 - System, apparatus, and method for asymmetrical beamforming with equal-power transmissions - Google Patents

Abstract

The present invention provides a plurality of embodiments for beamforming in an asymmetric wireless communication system 400 consisting of N _T transmit antennas 102 _i and N _R receive antennas 104 _j , where N _T > N _R ), the antennas ensure that the transmit power at each antenna is the same without any appreciable loss in performance. Techniques are also provided for selecting fewer beamforming vectors than frequency bins in an OFDM system.

Description

[0001] SYSTEM, APPARATUS, AND METHOD FOR ASYMMETRICAL BEAM FORMING WITH EQUAL-POWER TRANSMISSIONS FOR SAME-POWER TRANSMISSION [0002]

The present invention relates to asymmetric beam forming techniques in a wireless network to ensure that the transmit power across all antennas is the same.

A beamforming technique using singular-value-decomposition (SVD) of a channel matrix is a well known method for improving performance when multiple antennas are available. If the number of transmit and receive antennas is the same, the beamforming matrix is intended to make the power transmitted from each antenna equal. However, in many cases the number of transmit antennas N _T is greater than the number of receive antennas N _R. In this situation, using only a subset of the eigenvectors corresponding to the largest singular value would result in unequal transmit power across the antenna. This situation is undesirable because most transmit chains have limited peak power.

The present invention provides a number of embodiments of techniques for implementing beamforming techniques to ensure that the transmit power across all antennas is the same in an asymmetric system.

Beamforming from a complex transmit antenna to a complex receive antenna is a well known method of extracting channel diversity. It is known that using an eigenvector of the channel matrix for beamforming is an optimal strategy when the number of transmit antennas N _T is equal to the number of receive antennas N _R. For an asymmetric situation (N _T > N _R ), a commonly used method is to select an eigenvector corresponding to the largest eigenvalue as a beamforming vector. The problem with this approach is that it generates unequal transmit power from each antenna. This is a problem because in most cases the RF chain is limited in peak-power.

That is, when multiple RF front-ends are used in a transmitter with multiple antennas, it is a good practice to have each chain transmit the same power. The reason is that it is not easy to raise the power of one chain while lowering the power of another chain in order to keep the total transmit power constant because most RF amplifiers are limited by peak power. This is even more so for orthogonal frequency division multiplexing (OFDM) systems where the signal naturally has a large peak-to-average ratio and the RF amplifier has a back-off to keep them operating in a linear range.

Most currently used beamforming methods with more transmit antennas than receive antennas have unequal transmit power in each transmit chain. The problem of equal power transmission for the case of N _T X 1 of a single transmit stream has been solved (KK Mukkavilli, A. Sabharwal, E. Erkip and B. Aazhang, "On beamforming with finite rate feedback in multiple-antenna systems, &Quot; Grassmanian beamforming for multiple-input multiple-output wireless systems, "IEEE Trans. Inform. Theory, vol. 49, no. 10, pp. 2562--2579, Oct.2003, IEEE Trans. Inform. Theory, vol. 49, no. 10, pp. 2735--2747, Oct. 2003).

However, the problem for more than one transmission stream has not been solved.

The present invention provides several embodiments for asymmetric beamforming that ensure that the transmit power at each antenna is the same without appreciable loss in performance. Techniques are also provided for selecting less beamforming vectors than frequency bins in an OFDM system. This latter technique is useful in embodiments where the vector is fed back instead of assuming that the transmitter has channel awareness and can calculate the vector.

The preferred embodiment includes:

Technique 1: Brute force normalization;

Technique 2: Quantization to only ± 1 ± j values;

Technique 3: Optimization based on accident rate probability;

Technique 4: Hybrid optimization;

Technique 5: Frequency domain optimization

Provides technology that includes.

The present invention applies to open and closed loop systems, that is, the open-loop system of electrons has a transmitter that uses one of the techniques described above for estimating Q and adjusting Q with the recognition of the channel, The loop system causes the receiver to perform this operation.

1 illustrates an asymmetric communication system having a feedback channel;

Figure 2 illustrates a method for determining a beamforming matrix for a closed loop asymmetric communication system in accordance with the present invention.

3 illustrates a closed loop device for determining and feedbacking a beamforming matrix having the same power in an asymmetric communication system;

Figure 4 illustrates a modified asymmetric closed-loop communication system in accordance with the present invention;

Figure 5 illustrates the performance of various techniques for quantization;

In the following detailed description, reference is made to the accompanying drawings which show, by way of example only, a specific closed loop embodiment in which the invention may be practiced. Skilled artisans will appreciate that these embodiments are for illustrative purposes only and that the location and arrangement of the individual components disclosed herein may be modified without departing from the spirit and scope of the invention as embodied in the appended claims, The present invention is not intended to limit the scope of the present invention in any way. That is, the above detailed description should not be taken in any limiting sense, and the scope of the present invention is limited only by the appended claims and their equivalents, for example, the transmitter can estimate and adjust Q. In the drawings, like numbers refer generally to the same or like features.

The present invention provides a plurality of low complexity techniques for achieving the spatial diversity benefits provided by the combined transmit antenna beam forming technique and the multiple receive antenna combining technique. All of the beamforming techniques of the preferred embodiment of the present invention require channel information at the transmitter.

Figure 1 illustrates two wireless stations 101, 102, 103, 104 that can be part of a wireless local area network (WLAN) that includes a mobile station (laptop, personal digital assistant (PDA) 105 may be part of a wide area wireless network and may be a wireless personal area network Such stations 101 and 105 may be, for example, wireless, such as IEEE 802.11, Standard or any other standard, and the compliance is partial or complete. However, each of the wireless stations 101, 105 has a plurality of antennas, and in the present invention it is assumed that the number of antennas is asymmetric.

Given a transmit beam forming and receive combining multi-antenna system (open and closed loop) with N _T transmit antennas and N _R receive antennas, where an N _R data stream is transmitted and the beamforming matrix is defined as Q I suppose. The signal model is then given by:

Where n is a noise vector and the received vector r is an N _R X 1 vector and the channel matrix H 103 is N _R X N _T Wherein the beamforming matrix (Q) is N _T X N _R Matrix, and x is an N _R X 1 vector. The channel (H) is assumed to be perfectly known. The transmitted vector is y = Q x , which is the N _T X 1 vector. In an OFDM system, the signal model is repeated for each frequency bin. In the frequency selective channel, H and Q are different for each frequency bin.

1, a closed loop system 100 is assumed and current channel state information is transmitted between stations (STAs) 101 and 105 to reduce decoding complexity. Each of the STAs 101 and 105 includes a number of antennas N _T (102 _i ) and N _R (104 _j ), respectively, which form the system 100. The communication bandwidth used for this purpose is referred to as the " feedback bandwidth ", and in some preferred embodiments, a channel estimator 106 (e.g., a quadrature demodulator) that represents current channel state information by a beamforming matrix Q determined using singular value decomposition The feedback from the receiver 105 to the transmitter 101 via the feedback channel 107 is fed back. The transmitter 101 uses a beamforming matrix Q to transmit each outgoing signal on a multi-spatial channel.

이 단일 고유벡터(1,0)만을 갖는 경우}, P는 행렬이 역행렬을 가질 수 없으므로 A는 고유 분해(eigen decomposition)를 갖지 못한다. 그러나 A가 m이 n보다 큰 m X n 실수 행렬일 경우에, A는 소위, 방정식 A=UDV^T의 특이 값 분해를 이용하여 써질 수 있다. 여기서 U는 m X n 행렬이고 V는 n X n 스퀘어 행렬이며, 이 둘은 U^TU=V^TV=I 및 D가 n X n 대각선 행렬이 되도록 직교 열을 갖는다. 복소수 행렬(A)에 대해서, 상기 특이 값 분해는 방정식 A=U^HDV로의 분해이고, 여기서 U 및 V는 단위 행렬이고, U^H는 U의 켤레 전치 행렬(conjugate transpose)이고 D는 대각선 행렬이며, 이의 구성요소는 상기 초기 행렬의 특이 값이다. A가 복소수 행렬일 경우에, 양의 특이 값을 갖는 이러한 분해가 항상 존재한다.If the matrix of eigenvectors (P) of a given matrix (A) is not a square matrix {

Has only a single eigenvector (1, 0)}, P does not have eigen decomposition because A can not have an inverse matrix. However, if A is an m X n real matrix with m greater than n, then A can be written using so-called singular value decomposition of the equation A = UDV ^T. Where U is an m X n matrix and V is an n X n square matrix, both of which have orthogonal columns such that U ^T U = V ^T V = I and D are n X n diagonal matrices. For the complex matrices A, the singular value decomposition is decomposition into the equation A = U ^H DV, where U and V are unitary matrices, U ^H is a conjugate transpose of U and D is a diagonal matrix , Its component is the singular value of the initial matrix. When A is a complex matrix, this decomposition with positive singular values always exists.

Let H = USV ^H be the SVD decomposition of the channel matrix H. The optimal choice for Q is then Q = [ V ₁ V ₂ ... V _R ], where V _i is the I th column of the matrix V. The requirement that the power transmitted from each antenna be the same is a constraint that each row of the beamforming matrix Q has the same power. Since R e = T, since each eigenvector V _i is orthonormal, each component of the transmitted vector v has the same transmitted power. However, in the case of R <T, this is no longer true.

Figure 2 illustrates a method 200 in accordance with the present invention. In step 201, the channel H is estimated. In a preferred embodiment, a modified channel estimator / power equalizer / feedback device 300 is provided. However, if feedback is not required by the transmitter, only a modified channel estimator / power equalizer device is provided. In either case, the memory 301 is included in the device, and H (301.1) is stored in this memory in step 201. In step 202, a beamforming matrix (Q) 301.2 is determined (as described above) and stored in the memory 301. Next, in step 203, the beamforming matrix (Q) 301.2 is adjusted using one of the following techniques, each of which may be used to determine whether the transmitted vector has the same power And includes separate preferred embodiments. The adjusted beamforming matrix 301.3 is stored in the memory 301 of the device 300.

Technique 1: normalization of deterrent techniques.

Q = [ V ₁ V ₂ ... V _R ]. Then, each row of Q is normalized to be unit power. The resulting beamforming matrix ensures the same power component of v .

Technique 2: Quantization to only ± 1 ± j values.

Again, begin with Q as defined above. Then, Q1 = sign [Re (Q) + jsign (Im (Q))] is a beamforming matrix that not only has the same power component but also can have only one of the four values , So fewer bits are used for feedback.

Technique 3: Optimization based on accident rate probability.

The previously described technique for obtaining a beamforming matrix having the same power row does not include any optimality criterion. Assume that each element of Q is ± 1 ± j. Then, the criterion for selecting Q is to maximize | det (HQ) |. ^Since there is a probable Q matrix of 4 ^NTNR , the search for the deterrent scheme will be too complicated. The preferred simplification includes that (1) the phase of Q is not a problem, Q ₁₁ can be arbitrarily set to 1 + j and (2) search only those columns that are orthogonal to each other. Using the two preferred simplifications described above, the search space is dramatically reduced. For example, in the case of 4 X 2, there are only nine vectors that are orthogonal to the vector with entries ± 1 ± j. Therefore, the search space is reduced from 65536 to 64 * 9 = 576. In this way, there is no need for the SVD to be executed.

Technique 4: Hybrid optimization.

Technique 3 above still requires optimization over a number of possibilities. An additional simplification is to use technique 2 for the first vector, i. E., To quantize the first vector of the SVD matrix and then use technique 3 to determine another vector. For the case of 4 X 2, this requires the SVD implementation, followed by optimization over the nine possible choices.

Technique 5: Optimization across the frequency domain.

을 최대화하는 Q를 선택하는 것이다. 이 검색 공간은 이전과 같다. 또 다시, SVD는 전혀 요구되지 않는다.In the case where a single beamforming matrix is selected for p channel frequency bins,

Quot; Q &quot; This search space is the same as before. Again, SVD is not required at all.

Referring now to FIG. 2, a method for determining a beamforming matrix Q in a closed loop including a receiver 105 and feeding Q back to the transmitter 101 is illustrated. In step 201, the receiver estimates the channel state in a matrix H. Then, in step 202, the beamforming matrix Q is estimated from H (as described above). In step 203, any of the techniques 1 to 5 of the present invention are used to adjust the matrix Q so that the components have the same power and in step 204 the steered beamforming matrix is fed back to the transmitter.

FIG. 3 includes a memory 301 for storing data 301.1 related to the channel state matrix H according to the present invention, data 301.2 related to the initial beamforming matrix, and data associated with the adjusted beamforming matrix An apparatus 300 for channel estimation and feedback in a closed loop according to the present invention is illustrated. The apparatus 300 further includes a power equalizer component 302 that receives the received signal 303 for the channel H and generates a channel matrix H from the channel estimator module 302.1, And a channel estimator module 302.1 for storing it in memory 301 as matrix / data 301.1. The power equalizer component 302 further comprises a beamforming matrix adjustment module 302.2 that forms an initial beamforming matrix and adjusts the initial beamforming matrix according to one of the techniques pre- (203) and stores the data associated with the adjusted matrix (Q) in the memory (301) with adjusted beamforming matrix / data (301.3). Finally, the power equalizer component includes a feedback module 302. 3 that feeds back the adjusted beamforming matrix Q as a feedback signal 304 via a feedback channel 107 to the transmitter 101.

4 is a block diagram of an embodiment of the present invention that interfaces with a channel estimator / feedback device 300 configured in accordance with the present invention and includes at least one Loop asymmetric communication system 400 including a transmitter 101 and a receiver 105. The channel estimator / feedback device 300 estimates the channel and generates and stores the channel matrix H and associated data 301.1 in the memory and obtains an initial beamforming matrix 301.2 from the channel matrix H And adjusts the initial beamforming matrix according to a technique preselected in the techniques 1 to 5 of the present invention (203), and outputs the adjusted beamforming matrix (Q) 301.3 to the memory . Finally, the channel estimator / feedback apparatus 300 feeds back the adjusted beamforming matrix (Q) 304 to the transmitter 101 using the feedback channel 107 (204). As shown above, the communication system 400 may follow any or all of the communication standards, such as IEEE 802.11, for example, and may be part of any type of wireless communication network. The present invention is intended to apply to all asymmetric wireless communication networks / systems.

Figure 5 illustrates the performance of various techniques for quantizing a 4 X 2 system in a frequency-selective system using bit-interleaved coded modulation. It can be seen that this technique has a very small performance loss, especially at a higher rate (ratio 5/6 64QAM) compared to optimal beamforming with unequal transmit power.

While the present invention has been described in connection with a particular embodiment, that is, a closed loop, it should be understood that modifications and variations will occur to those skilled in the art, without departing from the spirit and scope of the invention as embodied in the appended claims. In particular, the transmitter may have the knowledge of the channel and perform the method without requiring any feedback from the receiver.

As described above, the present invention is available for asymmetric beamforming techniques in wireless networks to ensure that the transmit power across all antennas is the same.

Claims (16)

A method for asymmetric beamforming transmission of a vector over a wireless channel {(H) (103)}, And step (here, N _T> N _R> 0) to provide a -N _T transmit antennas (102 _i) and the N _R receive antennas (104 _j), the wireless communication system 400 having, - preselecting one of the adjustment techniques for obtaining a beamforming matrix having the same power row, - adjusting Q with a pre-selected adjustment technique such that each row of the beamforming matrix (Q) 301.2 has the same power, - transmitting the vectors as N _R data streams by the transmitter 101 through the channel {(H) (103)} using the adjusted beamforming matrix 301.3, The transmitted N _R data streams have the same power, The preselected adjustment technique consists of normalization of the suppression scheme to adjust the beamforming matrix, quantization with a value of ± 1 ± j (where j is greater than 0), optimization based on probability of accident rate, hybrid optimization and frequency domain optimization &Lt; / RTI &gt; A method for asymmetric beamforming transmission of a vector over a wireless channel. The method according to claim 1, Wherein the adjusting is performed at the transmitter (101) in accordance with a preselected brute force normalization in a pre-selecting adjustment technique. delete The method according to claim 1, Wherein the adjusting comprises: - estimating the channel (H) at a receiver (105) - performing the adjusting step at a receiver (300) - feedback (107) the adjusted beamforming matrix to the transmitter (101) &Lt; / RTI &gt; further comprising transmitting the asymmetric beamforming vector over a wireless channel. 5. The method of claim 4, And combining the transmitted and beamformed N _R data streams by the receiver (105). A method for asymmetric beamforming transmission of a vector over a wireless channel. A beam forming apparatus (300) for a multi-antenna system having N _T transmit antennas and N _R receive antennas with a relationship of N _T > N _R > 0, A memory 301 for storing information 301.1 to 301.3 for equalizing the beam forming power, Receiving the received signal 303, estimating the information to equalize the beamforming transmit power over the N _R data streams in accordance with a preselected equalization technique, and for storing information 301.1 to 301.3 in the memory Power equalizer component 302, The pre-selected equalization technique consists of normalization of the suppression scheme to adjust the beamforming matrix, quantization with a value of ± 1 ± j (where j is greater than 0), optimization based on probability of accident rate, hybrid optimization and frequency domain optimization &Lt; / RTI &gt; A beam forming apparatus for a multi-antenna system. delete The method according to claim 6, The memory 301 comprises channel state information 301.1, beamforming information 301.2 and adjusted beamforming information 301.3, The power equalizer component 302 includes a channel estimator module 302.1 for estimation and storage of the channel state information 301.1 in the memory and a channel estimator module 302.1 for determining and storing in the memory of the beamforming information 301.2 And for adjusting and storing in the memory the adjusted beamforming information (301.3) to equalize the transmit power of the N _R data streams to the adjusted beamforming information (301.3) (302.2). &Lt; / RTI &gt; 9. The method of claim 8, Wherein the power equalizer component (302) further comprises a feedback module (302.3) for providing a feedback signal (303) comprising the adjusted beamforming information (301.3). A transmit beamforming and receive combining multi-antenna system (400) comprising: - one or more receivers (105) comprising N _R receive antennas (104 _j ), where N _T > N _R &gt; And N _R receive antennas (104 _j) into the N _R data streams transmitted beam forming the N _T transmit antennas (102 _i) at least one transmitter (101) comprising for the, - In accordance with a pre-selected equalization technique, an adjusted beamforming matrix to be used by one or more transmitters 101 and a beamforming matrix (e.g., a beamforming matrix) to be used in the memory 301 are used to equalize the beamforming transmit power over the transmitted N _R data streams. (300) for providing a storage of a channel estimator (301.3), wherein the channel estimator / The pre-selected equalization techniques include normalization of the suppression scheme to output the adjusted beamforming matrix, quantization with a value of ± 1 j (where j is greater than 0), optimization based on probability of accident rate, hybrid optimization and frequency domain optimization &Lt; / RTI &gt; Transmit beamforming and receive combining multiple-antenna system. delete 11. The method of claim 10, Wherein at least one channel estimator / power equalizer device (300) is operably coupled to one or more transmitters (101), and the pre-selected equalization technique is a normalization of the suppression technique. 11. The method of claim 10, At least one channel estimator / power equalizer device 300 is one or more of the further include the associated feedback module (302.3) operable to receiver 105, the feedback module (302.3) is over the N _R of the data stream the transmitted Configured to provide a feedback signal (303) comprising an adjusted beamforming matrix (301.3) to be fed back to the transmitter (101) via a feedback channel (107) for use to equalize the beamforming transmit power, Beam forming and receiving combining multi-antenna system. A beam forming transmitter apparatus comprising: And N _T> N _R> N _R receive antennas (104 _j) N _T transmit antennas for the transmission beamforming vector of a channel through the N _R data streams to the (102 _i) having a relationship of 0, - A power equalizer component 302 that determines an estimate of the channel and a beamforming matrix from which the beamforming matrices are adjusted to equalize the beamforming transmit power over the N _R data streams in accordance with a preselected equalization technique )Wow, - a transmitter using an adjusted beamforming matrix to transmit the vector as N _R data streams having the same power, The pre-selected equalization techniques include normalization of the suppression scheme to output the adjusted beamforming matrix, quantization with a value of ± 1 j (where j is greater than 0), optimization based on probability of accident rate, hybrid optimization and frequency domain optimization &Lt; / RTI &gt; Beam forming transmitter device. A combined receiver device, - N _T> N _R> N for receiving and combining the beamforming transmission by a transmitter 101 having N _R streams of data channels of N _T the vector transmission antenna via (102 _i) having a zero relationship _R and receive antennas (104 _j), - providing the adjusted beamforming matrix to be fed back for use by the transmitter 101 making the beamforming transmit power over the transmitted N _R data streams equal to the channel estimator / An equalizer / feedback device 300, The pre-selected equalization techniques include normalization of the suppression technique to generate an adjusted beamforming matrix, quantization (where j is greater than 0) to ± 1 ± j values, optimization based on accident rate probabilities, hybrid optimization and frequency domain optimization &Lt; / RTI &gt; Coupled receiver device. delete

Images